One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
On 14/11/2026 11:10, Don Y wrote:
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Larger drives in the 20-40Tbyte range have much closer ratios.
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Don Y <blockedofcourse@foo.invalid> wrote:But what manufacturing issue determines 1.0T vs 1.1T?
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Typically there's a certain number of bits you can store per platter. eg let's say it's 1T per side.
One read head on one platter gets you 1T
Two read heads on both sides of one platter gets you 2T
Two platters with four heads gets you 4T
Four platters with eight heads gets you 8T
etc etc.
It's not worth retooling for a new generation for only a minor bump to 1.1T per platter, so these models hang around until somebody can manage 1.5T per platter, then we get 1.5T/3T/6T/12T. Then the next bump is to 2T per
platter so 2T/4T/8T/16T. Rinse and repeat.
Apart from the largest sizes, there's always an upgrade path within the generation: ie instead of waiting for the new generation to upgrade from 1T to 1.5T, you can upgrade to the next step of 2T in the current generation. This is much lower cost for the vendor (no retooling, just the extra platters/heads).
On 11/14/2026 5:37 AM, Theo wrote:
Apart from the largest sizes, there's always an upgrade path within the generation: ie instead of waiting for the new generation to upgrade from 1T to 1.5T, you can upgrade to the next step of 2T in the current generation. This is much lower cost for the vendor (no retooling, just the extra platters/heads).But what manufacturing issue determines 1.0T vs 1.1T?
Magnetic domains don't have physical boundaries that
have to be reshaped.
I.e., why "settle" on 1T (per platter/device/etc) and not 1.2T?
Platter dimensions aren't being tweeked so why the "coincidence"
in the capacities available?
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Don Y <blockedofcourse@foo.invalid> wrote:
On 11/14/2026 5:37 AM, Theo wrote:
But what manufacturing issue determines 1.0T vs 1.1T?
Apart from the largest sizes, there's always an upgrade path within the
generation: ie instead of waiting for the new generation to upgrade from 1T >>> to 1.5T, you can upgrade to the next step of 2T in the current generation. >>> This is much lower cost for the vendor (no retooling, just the extra
platters/heads).
Magnetic domains don't have physical boundaries that
have to be reshaped.
I.e., why "settle" on 1T (per platter/device/etc) and not 1.2T?
Platter dimensions aren't being tweeked so why the "coincidence"
in the capacities available?
You need a certain amount of 'raw' data for servo information and for blocks reallocated due to errors. Since round numbers are better for marketing (1T not 1.134T) I expect those hidden areas can vary in size so as to make the advertised capacity a round number.
For operators of multi-disc arrays, having a non-round size is a bug not a feature. If the 1.134T drive in an array fails, you need another 1.134TYes, hence the creation of DSO.
or larger drive to replace it. The competitor's 1.132T drive won't do.
Yes. The next version has to be twice as good as the previous version and
1TB (I'm surprised it's not 1000GB) is twice as good as 500GB. Isn't it?
Also it's perfectly reasonable to charge twice the price for twice the storage
capacity. Isn't it?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Amusingly, I have seen 1T drives and 1000GB drives from the same vendor
(with the same capacity); marketeering?
On 11/14/2026 9:45 AM, Edward Rawde wrote:
Yes. The next version has to be twice as good as the previous version and
1TB (I'm surprised it's not 1000GB) is twice as good as 500GB. Isn't it?
Often, such models are available concurrently so its a question of
what the company decided to "push".
Also it's perfectly reasonable to charge twice the price for twice the storage
capacity. Isn't it?
IME (admittedly sorely out of date as I haven't paid for kit in decades),
the cost of storage media has dropped (price per bit/spindle) considerably.
I recall paying $10K for ten 4G (that's a G, not a T) disks, ages ago.
And, the media is considerably more durable -- I've seen drives with ~100K PoH
and *no* grown defects!
Remember when drives were mounted on overdesigned "shock mounts" out of
fear the slightest jarring would crash the disk?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
On Sat, 14 Nov 2026 04:10:20 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Certainly. Customers and end users prefer easily digestible whole
numbers. Anything that require math to understand, such as decimal arithmetic, turns off the GUM (great unwashed masses). Western Dismal
makes it even simpler for buyers using color coding.
Large numbers sell better than small numbers. There's about a 25%
increase in drive capacity when changing from CMR (conventional
magnetic recording) to SMR (shingled magnetic recording): <https://buffaloamericas.com/resources/cmr-vs-smr-hard-drives-in-network-attached-storage-nas-msp>
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long
lead times are causing buyers to buy almost anything that spins. The increases in HDD capacity are causing a drop in the number of drives
shipped: <https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of
minor differences in price.
On Sat, 14 Nov 2026 10:10:33 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
Amusingly, I have seen 1T drives and 1000GB drives from the same vendor
(with the same capacity); marketeering?
Is that decimal or binary TB?
"Why does my hard drive report less capacity than indicated on the
drive's label?" <https://www.seagate.com/support/kb/why-does-my-hard-drive-report-less-capacity-than-indicated-on-the-drives-label-172191en/>
"Storage capacity measurement standards" <https://www.seagate.com/support/kb/storage-capacity-measurement-standards-002046en/>
Occam's Razor. The simplest explanation is the manufacturer ran out
of space on the label. 1T occupies less printing area on the label
than 1000GB.
There was a time, when HD manufactures provided labels with everything
EXCEPT the drive capacity. These drives had the heads, tracks,
cylinders, platters, sectors/track, but not total capacity.
On Sat, 14 Nov 2026 04:10:20 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Certainly. Customers and end users prefer easily digestible whole
numbers. Anything that require math to understand, such as decimal arithmetic, turns off the GUM (great unwashed masses). Western Dismal
makes it even simpler for buyers using color coding.
Large numbers sell better than small numbers. There's about a 25%
increase in drive capacity when changing from CMR (conventional
magnetic recording) to SMR (shingled magnetic recording): <https://buffaloamericas.com/resources/cmr-vs-smr-hard-drives-in-network-attached-storage-nas-msp>
But, that doesn't address historical trends.Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long
lead times are causing buyers to buy almost anything that spins. The increases in HDD capacity are causing a drop in the number of drives
shipped: <https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of
minor differences in price.
"Don Y" <blockedofcourse@foo.invalid> wrote in message news:11da5ao$2489f$1@dont-email.me...
On 11/14/2026 9:45 AM, Edward Rawde wrote:
Yes. The next version has to be twice as good as the previous version and >>> 1TB (I'm surprised it's not 1000GB) is twice as good as 500GB. Isn't it?
Often, such models are available concurrently so its a question of
what the company decided to "push".
Sure but the buyer will see something twice as big.
And twice as big is twice as good. Isn't it?
Also it's perfectly reasonable to charge twice the price for twice the storage
capacity. Isn't it?
IME (admittedly sorely out of date as I haven't paid for kit in decades),
the cost of storage media has dropped (price per bit/spindle) considerably. >> I recall paying $10K for ten 4G (that's a G, not a T) disks, ages ago.
It seems to have reached a turning point and is now increasing again. Probably because you shouldn't need storage at home.
It should be in the "cloud". Shouldn't it?
And, the media is considerably more durable -- I've seen drives with ~100K PoH
and *no* grown defects!
Remember when drives were mounted on overdesigned "shock mounts" out of
fear the slightest jarring would crash the disk?
I remember disks which had to be pulled out of a unit which looked like
a top loading washing machine.
There are some advantages to having a small number of well defined
drive sizes.
Recently I wanted to make an image backup of the 2Tbyte SSD (*) in my
laptop PC. I have lots of ex data-centre drives, many of which
are "2Tbyte".
It turned out that the SSD capacity was EXACTLY the same number of
bytes as the spinning disks,so the image fitted perfectly.
* I bought the SSD before prices became stupidly expensive.
John
On 11/14/2026 11:23 AM, Jeff Liebermann wrote:
On Sat, 14 Nov 2026 04:10:20 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Certainly. Customers and end users prefer easily digestible whole
numbers. Anything that require math to understand, such as decimal
arithmetic, turns off the GUM (great unwashed masses). Western Dismal
makes it even simpler for buyers using color coding.
The coloration indicates other drive characteristics beyond
capacity (Red drives for NAS use, Blue consumer, Black enterprise,
green power saving, gold... )
Large numbers sell better than small numbers. There's about a 25%
increase in drive capacity when changing from CMR (conventional
magnetic recording) to SMR (shingled magnetic recording):
<https://buffaloamericas.com/resources/cmr-vs-smr-hard-drives-in-network-attached-storage-nas-msp>
Shingled recording takes a performance hit. (and, likely, a
data reliability hit). It is really only useful for "write seldom" >applications (like large consumer external disks acting as archival
storage).
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long
lead times are causing buyers to buy almost anything that spins. The
increases in HDD capacity are causing a drop in the number of drives
shipped:
<https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of
minor differences in price.
But, that doesn't address historical trends.
On 11/14/2026 12:07 PM, Edward Rawde wrote:
"Don Y" <blockedofcourse@foo.invalid> wrote in message news:11da5ao$2489f$1@dont-email.me...
On 11/14/2026 9:45 AM, Edward Rawde wrote:
Yes. The next version has to be twice as good as the previous version and >>>> 1TB (I'm surprised it's not 1000GB) is twice as good as 500GB. Isn't it? >>>Often, such models are available concurrently so its a question of
what the company decided to "push".
Sure but the buyer will see something twice as big.
And twice as big is twice as good. Isn't it?
The amusing question is why PC vendors don't opt for larger
drives in their offerings
. At their buying quantities, I
would imagine doubling the size would cost next to nothing.
Also it's perfectly reasonable to charge twice the price for twice the storage
capacity. Isn't it?
IME (admittedly sorely out of date as I haven't paid for kit in decades), >>> the cost of storage media has dropped (price per bit/spindle) considerably. >>> I recall paying $10K for ten 4G (that's a G, not a T) disks, ages ago.
It seems to have reached a turning point and is now increasing again.
Probably because you shouldn't need storage at home.
It should be in the "cloud". Shouldn't it?
It could be that SSDs are now replacing rust drives.
And, the media is considerably more durable -- I've seen drives with ~100K PoH
and *no* grown defects!
Remember when drives were mounted on overdesigned "shock mounts" out of
fear the slightest jarring would crash the disk?
I remember disks which had to be pulled out of a unit which looked like
a top loading washing machine.
RK07?
I used to have an "in" at DECs Maynard facility -- free reign
of the "scrap" room. I recall seeing a pair of these where the disk (pack) had crashed "spectacularly". It's one thing to rotate a tiny platter at 15KRPM but entirely different to rotate a massive *pack*!
[IIRC, someone made a 3D display out of one of these by mounting a plane of LEDs on it and sequencing them as it revolved]
I had a *fixed* disk drive (RS03, IIRC) that had 512KB (that's a 'K') of capacity effectively WORD ADDRESSABLE -- at about 500KHz. It weighed
a bit over 100 pounds and would dim the lights when starting.
*AND*, I considered myself PRIVILEGED to have such a thing! <rolls eyes>
One thing I learned, early on, was never to buy things as they were
obsoleted too quickly!
On 11/14/2026 5:37 AM, Theo wrote:
Don Y <blockedofcourse@foo.invalid> wrote:But what manufacturing issue determines 1.0T vs 1.1T?
One can understand why semiconductor memory tends to follow certain
size patterns.
But, there is no similar constraint on spinning rust (or, SSDs,
for that matter).
So, why do disk sizes seem to follow a particular pattern?
250G, 500G, 1T, 2T, 4T, 8T, etc.?
Yes, there are some oddballs in the mix -- 320G, 640G, 750G, 1.5T,
3T, 14T, etc. -- but, for the most part, the "doubling" ratio seems
most common.
Marketing?
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Typically there's a certain number of bits you can store per platter.-a eg >> let's say it's 1T per side.
One read head on one platter gets you 1T
Two read heads on both sides of one platter gets you 2T
Two platters with four heads gets you 4T
Four platters with eight heads gets you 8T
etc etc.
It's not worth retooling for a new generation for only a minor bump to
1.1T
per platter, so these models hang around until somebody can manage
1.5T per
platter, then we get 1.5T/3T/6T/12T.-a Then the next bump is to 2T per
platter so 2T/4T/8T/16T. Rinse and repeat.
Apart from the largest sizes, there's always an upgrade path within the
generation: ie instead of waiting for the new generation to upgrade
from 1T
to 1.5T, you can upgrade to the next step of 2T in the current
generation.
This is much lower cost for the vendor (no retooling, just the extra
platters/heads).
Magnetic domains don't have physical boundaries that
have to be reshaped.
I.e., why "settle" on 1T (per platter/device/etc) and not 1.2T?
Platter dimensions aren't being tweeked so why the "coincidence"
in the capacities available?
I.e., why "settle" on 1T (per platter/device/etc) and not 1.2T?
Platter dimensions aren't being tweeked so why the "coincidence"
in the capacities available?
Available technology.
You develop a technology that allows a certain density of data per inch. Well,
the diameter of the plate is given, so the size in bytes of a plate is thus given, not arbitrary.
On 11/14/2026 12:15 PM, John R Walliker wrote:
There are some advantages to having a small number of well defined
drive sizes.
Recently I wanted to make an image backup of the 2Tbyte SSD (*) in my
laptop PC. I have lots of ex data-centre drives, many of which
are "2Tbyte".
It turned out that the SSD capacity was EXACTLY the same number of
bytes as the spinning disks,so the image fitted perfectly.
I store images in compressed form -- along with a hash to reassure me
of their integrity.-a A smart image creator doesn't bother storing
unused portions of the medium (it has to be filesystem-aware to do so).
E.g., I store several iterations of the "build process" when creating
a new Windows box -- on a 250G disk even thought he system may
have a 1T drive within.-a (this is a win if I botch the install and
want to rollback to "just after the drivers were installed" or
"just before installing the drivers" or...)
One can usually install an image onto a larger medium and "enlarge"
the filesystem to fit.
Depending on what is where, you can also often SHRINK a filesystem
to fit on a smaller medium.
* I bought the SSD before prices became stupidly expensive.
John
On 14/11/2026 20:40, Don Y wrote:Clonezilla is a handy tool to have in your kit. Almost foolproof
On 11/14/2026 12:15 PM, John R Walliker wrote:
There are some advantages to having a small number of well defined
drive sizes.
Recently I wanted to make an image backup of the 2Tbyte SSD (*) in my
laptop PC. I have lots of ex data-centre drives, many of which
are "2Tbyte".
It turned out that the SSD capacity was EXACTLY the same number of
bytes as the spinning disks,so the image fitted perfectly.
I store images in compressed form -- along with a hash to reassure me
of their integrity.-a A smart image creator doesn't bother storing
unused portions of the medium (it has to be filesystem-aware to do so).
I don't use smart image creators.-a I just use dd.-a Then sha256sum on the original and the copy.
Sure but the buyer will see something twice as big.
And twice as big is twice as good. Isn't it?
The amusing question is why PC vendors don't opt for larger
drives in their offerings
Marketing again.
Drive? What's that? Is that the memory where my pictures are stored?
Or is that the cloud?
Also it's perfectly reasonable to charge twice the price for twice the storage
capacity. Isn't it?
IME (admittedly sorely out of date as I haven't paid for kit in decades), >>>> the cost of storage media has dropped (price per bit/spindle) considerably.
I recall paying $10K for ten 4G (that's a G, not a T) disks, ages ago.
It seems to have reached a turning point and is now increasing again.
Probably because you shouldn't need storage at home.
It should be in the "cloud". Shouldn't it?
It could be that SSDs are now replacing rust drives.
They are, but I expect that data centres want a lot of them.
The issue was disk capacities. Historically, the same rough "doubling"Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long
lead times are causing buyers to buy almost anything that spins. The
increases in HDD capacity are causing a drop in the number of drives
shipped:
<https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of
minor differences in price.
But, that doesn't address historical trends.
I didn't want to post every graph I found showing sales by unit,
capacity, market segment, withdrawn for internal use, refurbished,
etc. Which historical trend would you like to see addressed where I
failed to address my observation that a volatile and demand driven
market is a bad time to compete on selling price?
On 11/14/2026 2:29 PM, Carlos E.R. wrote:
I.e., why "settle" on 1T (per platter/device/etc) and not 1.2T?
Platter dimensions aren't being tweeked so why the "coincidence"
in the capacities available?
Available technology.
You develop a technology that allows a certain density of data per
inch. Well, the diameter of the plate is given, so the size in bytes
of a plate is thus given, not arbitrary.
But, just so happens to end up a "nice number"?
On 11/14/2026 3:58 PM, John R Walliker wrote:
On 14/11/2026 20:40, Don Y wrote:
I don't use smart image creators.a I just use dd.a Then sha256sum on the
original and the copy.
Clonezilla is a handy tool to have in your kit. Almost foolproof
in operations (once you've been through it a time or two).
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long >>>> lead times are causing buyers to buy almost anything that spins. The
increases in HDD capacity are causing a drop in the number of drives
shipped:
<https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of >>>> minor differences in price.
But, that doesn't address historical trends.
I didn't want to post every graph I found showing sales by unit,
capacity, market segment, withdrawn for internal use, refurbished,
etc. Which historical trend would you like to see addressed where I
failed to address my observation that a volatile and demand driven
market is a bad time to compete on selling price?
The issue was disk capacities. Historically, the same rough "doubling"
of capacities has been the trend.
I recall 4G drives. Then 9G. 18G, 36G, 72G, 146G, 300G, 600G, 500G,
1T, 2T, 4T, 8T, etc.
No 19G or 310G drives. No 7T.
Regardless of price or quantities sold.
On Sat, 14 Nov 2026 16:46:20 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
On 11/14/2026 3:58 PM, John R Walliker wrote:
On 14/11/2026 20:40, Don Y wrote:
I don't use smart image creators.-a I just use dd.-a Then sha256sum on the >>> original and the copy.
Clonezilla is a handy tool to have in your kit. Almost foolproof
in operations (once you've been through it a time or two).
A GUI (graphical user interface) should make backups and restores
easier:
<https://rescuezilla.com>
On Sat, 14 Nov 2026 17:12:37 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long >>>>> lead times are causing buyers to buy almost anything that spins. The >>>>> increases in HDD capacity are causing a drop in the number of drives >>>>> shipped:
<https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of >>>>> minor differences in price.
But, that doesn't address historical trends.
I didn't want to post every graph I found showing sales by unit,
capacity, market segment, withdrawn for internal use, refurbished,
etc. Which historical trend would you like to see addressed where I
failed to address my observation that a volatile and demand driven
market is a bad time to compete on selling price?
The issue was disk capacities. Historically, the same rough "doubling"
of capacities has been the trend.
I asked for the "historical trend" which you claimed that my comments
did not "address".
Instead, you provided a disk size "issue" which
could be explained as an exercise in rounding and truncating long
strings of numbers. When one has a drive with a trillion bits, a few
missing or surplus bits are of little importance.
I recall 4G drives. Then 9G. 18G, 36G, 72G, 146G, 300G, 600G, 500G,
1T, 2T, 4T, 8T, etc.
No 19G or 310G drives. No 7T.
Regardless of price or quantities sold.
Are you suggesting that drive capacity should be selected by
intelligent design instead of natural selection?
Intelligent design
might yield a few cool interesting patterns and magical numbers.
However, natural selection simple scans the platters for errors, fills
them with some number of bits that the platters can safely hold, saves
some bits for error correction, and reserves some bits for wear
leveling and uncorrectable errors. The idea is to maximize the number
of usable bits over each platter. Such a scheme is unlikely to
produce your short and convenient disk capacity numbers.
Every drive manufacturer makes it clear that a terabyte is 1000 billion
bytes.
Don Y <blockedofcourse@foo.invalid> wrote:Being an american, it is the former. 1000 GB
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion
bytes.
Is that an American billion (billiard) or a real English billion (a bi-million)?
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion >>bytes.
Is that an American billion (billiard) or a real English billion (a >bi-million)?
On Sun, 15 Nov 2026 13:37:45 +0000, liz@poppyrecords.invalid.invalid
(Liz Tuddenham) wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion
bytes.
Is that an American billion (billiard) or a real English billion (a
bi-million)?
Thousand, million, billion, trillion, quadrillion.
Steps of 1e3, like the SI prefixes K M G T P.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
On 11/15/2026 11:43 PM, john larkin wrote:
On Sun, 15 Nov 2026 13:37:45 +0000, liz@poppyrecords.invalid.invalid
(Liz Tuddenham) wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion >>>> bytes.
Is that an American billion (billiard) or a real English billion (a
bi-million)?
Thousand, million, billion, trillion, quadrillion.
Steps of 1e3, like the SI prefixes K M G T P.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
And never mind the older generation of Brits who put "milliard" in there >somewhere.
On Sun, 15 Nov 2026 23:50:47 +0800, Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> wrote:
On 11/15/2026 11:43 PM, john larkin wrote:
On Sun, 15 Nov 2026 13:37:45 +0000, liz@poppyrecords.invalid.invalid
(Liz Tuddenham) wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion >>>>> bytes.
Is that an American billion (billiard) or a real English billion (a
bi-million)?
Thousand, million, billion, trillion, quadrillion.
Steps of 1e3, like the SI prefixes K M G T P.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
And never mind the older generation of Brits who put "milliard" in there
somewhere.
That sounds like something bad to eat, like marmite.
We have wonderful restaurants here in San Francisco. Italian, French, cajun-creole, bbq, thai, nepalese, mandarin, cantonese, mexican,
portugese, eritrean, yucatan, pizza, all sorts of stuff. We like some neighborhood places like Regent Thai and United Dumplings and Go Duck Yourself.
But there are no British or Irish or German restaurants. There is
Cinderella, a Russian bakery.
I think there is one kosher restaurant. It's awful.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
On 11/16/2026 12:39 AM, john larkin wrote:
On Sun, 15 Nov 2026 23:50:47 +0800, Johann 'Myrkraverk' Oskarsson
<johann@myrkraverk.invalid> wrote:
On 11/15/2026 11:43 PM, john larkin wrote:
On Sun, 15 Nov 2026 13:37:45 +0000, liz@poppyrecords.invalid.invalid
(Liz Tuddenham) wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion >>>>>> bytes.
Is that an American billion (billiard) or a real English billion (a
bi-million)?
Thousand, million, billion, trillion, quadrillion.
Steps of 1e3, like the SI prefixes K M G T P.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
And never mind the older generation of Brits who put "milliard" in there >>> somewhere.
That sounds like something bad to eat, like marmite.
We have wonderful restaurants here in San Francisco. Italian, French,
cajun-creole, bbq, thai, nepalese, mandarin, cantonese, mexican,
portugese, eritrean, yucatan, pizza, all sorts of stuff. We like some
neighborhood places like Regent Thai and United Dumplings and Go Duck
Yourself.
But there are no British or Irish or German restaurants. There is
Cinderella, a Russian bakery.
I think there is one kosher restaurant. It's awful.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
We have a German restaurant called Brotzeit. It's a chain with a few >outlets, but it's far from common; unlike Texas Roadhouse and Fridays.
On Sun, 15 Nov 2026 13:37:45 +0000, liz@poppyrecords.invalid.invalidITYM k M G T P.
(Liz Tuddenham) wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion
bytes.
Is that an American billion (billiard) or a real English billion (a
bi-million)?
Thousand, million, billion, trillion, quadrillion.
Steps of 1e3, like the SI prefixes K M G T P.
On 11/14/2026 8:03 PM, Jeff Liebermann wrote:
On Sat, 14 Nov 2026 17:12:37 -0700, Don Y
<blockedofcourse@foo.invalid> wrote:
Why don't we see manufacturers trying to nickel-and-dime each
other -- 2.1T, 4.4T, etc.
Rotating rust drives are in demand at AI data centers (due to a
shortage of SSD drives). HDD volume and margins are decreasing. Long >>>>>> lead times are causing buyers to buy almost anything that spins. The >>>>>> increases in HDD capacity are causing a drop in the number of drives >>>>>> shipped:
<https://www.statista.com/statistics/398951/global-shipment-figures-for-hard-disk-drives/>
An uncertain market is not a good time to be competing on the basis of >>>>>> minor differences in price.
But, that doesn't address historical trends.
I didn't want to post every graph I found showing sales by unit,
capacity, market segment, withdrawn for internal use, refurbished,
etc. Which historical trend would you like to see addressed where I
failed to address my observation that a volatile and demand driven
market is a bad time to compete on selling price?
The issue was disk capacities. Historically, the same rough "doubling"
of capacities has been the trend.
I asked for the "historical trend" which you claimed that my comments
did not "address".
The thread is about disk capacities (sizes).
So, HISTORICALLY, I
contend that the 2X pattern has been in place for decades. I cited
disk sizes of volumes I have personally owned and invite a discussion
of other "oddball" sizes and why those saw a market share.
Instead, you provided a disk size "issue" which
could be explained as an exercise in rounding and truncating long
strings of numbers. When one has a drive with a trillion bits, a few
missing or surplus bits are of little importance.
I recall 4G drives. Then 9G. 18G, 36G, 72G, 146G, 300G, 600G, 500G,
1T, 2T, 4T, 8T, etc.
No 19G or 310G drives. No 7T.
Regardless of price or quantities sold.
Are you suggesting that drive capacity should be selected by
intelligent design instead of natural selection?
No. I asked why the sizes always (including historically) seem
to follow a particular pattern/progression.
Why 5TB drives
aren't common, considering they have some additional appeal
beyond 4TB drives (presumably at a lower price point than 8TB)
The RS03 became the RS04 by making use of the "other side"
of the platter. So, you expected its capacity to be twice that
of the '03.
I don't see that to be the case with modern drives -- the--
"back side" is already in use. And, there is a limit
as to how many additional platters you can squeeze into
the case.
Intelligent design
might yield a few cool interesting patterns and magical numbers.
However, natural selection simple scans the platters for errors, fills
them with some number of bits that the platters can safely hold, saves
some bits for error correction, and reserves some bits for wear
leveling and uncorrectable errors. The idea is to maximize the number
of usable bits over each platter. Such a scheme is unlikely to
produce your short and convenient disk capacity numbers.
On Mon, 16 Nov 2026 00:54:19 +0800, Johann 'Myrkraverk' Oskarsson <johann@myrkraverk.invalid> wrote:
On 11/16/2026 12:39 AM, john larkin wrote:
On Sun, 15 Nov 2026 23:50:47 +0800, Johann 'Myrkraverk' Oskarsson
<johann@myrkraverk.invalid> wrote:
On 11/15/2026 11:43 PM, john larkin wrote:
On Sun, 15 Nov 2026 13:37:45 +0000, liz@poppyrecords.invalid.invalid >>>>> (Liz Tuddenham) wrote:
Don Y <blockedofcourse@foo.invalid> wrote:
[...]
Every drive manufacturer makes it clear that a terabyte is 1000 billion >>>>>>> bytes.
Is that an American billion (billiard) or a real English billion (a >>>>>> bi-million)?
Thousand, million, billion, trillion, quadrillion.
Steps of 1e3, like the SI prefixes K M G T P.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
And never mind the older generation of Brits who put "milliard" in there >>>> somewhere.
That sounds like something bad to eat, like marmite.
We have wonderful restaurants here in San Francisco. Italian, French,
cajun-creole, bbq, thai, nepalese, mandarin, cantonese, mexican,
portugese, eritrean, yucatan, pizza, all sorts of stuff. We like some
neighborhood places like Regent Thai and United Dumplings and Go Duck
Yourself.
But there are no British or Irish or German restaurants. There is
Cinderella, a Russian bakery.
I think there is one kosher restaurant. It's awful.
John Larkin
Highland Tech Glen Canyon Design Center
Lunatic Fringe Electronics
We have a German restaurant called Brotzeit. It's a chain with a few
outlets, but it's far from common; unlike Texas Roadhouse and Fridays.
Is the Roadhouse any good? There's one east of here on I80.
I remember a German restaurant in NYC that was good. Luchows. Roast
goose. It's closed now.
So, HISTORICALLY, I
contend that the 2X pattern has been in place for decades. I cited
disk sizes of volumes I have personally owned and invite a discussion
of other "oddball" sizes and why those saw a market share.
If you want to specify a rate of increase, you need to also specify
the rate increase over a specific time period. 2X is not sufficient.
From this graph, it was something like 10x every 5 years: <https://www.reddit.com/r/DataHoarder/comments/17u8c5q/the_7_orders_of_hdd_magnitude_1980_present/>
That was the increase from 1980 to 2005. Since then, the rate of
increase has decreased to about 2x every 5 years. I could not find
the source of that data used to create the graph. The graph also
indicate the changes in drive form factor. Kinda useless, but all I
could find. Ok, now where's YOUR data.
Instead, you provided a disk size "issue" which
could be explained as an exercise in rounding and truncating long
strings of numbers. When one has a drive with a trillion bits, a few
missing or surplus bits are of little importance.
I recall 4G drives. Then 9G. 18G, 36G, 72G, 146G, 300G, 600G, 500G,
1T, 2T, 4T, 8T, etc.
No 19G or 310G drives. No 7T.
Regardless of price or quantities sold.
Are you suggesting that drive capacity should be selected by
intelligent design instead of natural selection?
No. I asked why the sizes always (including historically) seem
to follow a particular pattern/progression.
Yes, and I asked you to provide evidence of such a trend.
Why 5TB drives
aren't common, considering they have some additional appeal
beyond 4TB drives (presumably at a lower price point than 8TB)
My guess(tm) is that major improvements in HDD form factor, oxide formulation, media density, platter RPM, head flying height, data
encoding, error recovery, mergers/acquisitions, etc are what drove the increases in drive capacity. I would not expect these to occur in an
orderly sequence, which might explain why drive capacity also did not increase in an orderly sequence.
The RS03 became the RS04 by making use of the "other side"
of the platter. So, you expected its capacity to be twice that
of the '03.
That was a DEC drive used in the PDP-10 in the late 1960's: <https://gunkies.org/wiki/RS03/04_disk_drive>
A more recent example might be more helpful.
You wrote "So, you expected its capacity to be twice...". I wrote no
such thing. You're putting words into my mouth and then complaining
that I said something wrong. I did not predict a doubling of capacity because of one major innovation. My guess(tm) is you're extrapolating
from the doubling in floppy disk capacity, which did happen when
double sided floppy disks were introduced. Also, using both sides of
a disk can double the capacity, but only once because there are only 2
sides to a platter. The capacity graph would show a 2X jump only once
then continue rising at the previous rate (assuming all other things
remained constant).
If using both sides of a platter was the only improvement in HDD drive construction, then I would expect to see a corresponding increase in
drive capacity. However, if were combined with other changes, the improvements in capacity would be a combination of these changes and
not increase at a constant 2X rate.
I don't see that to be the case with modern drives -- the
"back side" is already in use. And, there is a limit
as to how many additional platters you can squeeze into
the case.
Intelligent design
might yield a few cool interesting patterns and magical numbers.
However, natural selection simple scans the platters for errors, fills
them with some number of bits that the platters can safely hold, saves
some bits for error correction, and reserves some bits for wear
leveling and uncorrectable errors. The idea is to maximize the number
of usable bits over each platter. Such a scheme is unlikely to
produce your short and convenient disk capacity numbers.
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 76 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 56:23:43 |
| Calls: | 1,265 |
| Calls today: | 1 |
| Files: | 1,362 |
| D/L today: |
3 files (2,684K bytes) |
| Messages: | 285,806 |